Japanese public-domain books from the Aozora Bunko archive, cleaned and formatted for machine learning tasks. The data includes full-text content and metadata for literary works, processed to remove complex formatting tags like ruby characters.
Use Cases
- Fine-tune Japanese language models using the cleaned text column to capture literary and historical linguistic patterns.
- Conduct stylometric analysis to identify unique writing styles using the author and text fields.
- Train text-to-speech systems using the cleaned prose in the text column for high-quality literary narration.
Strengths
- Extracted from the Aozora Bunko repository, the primary digital archive for Japanese public-domain literature.
- Processed using the aozorabunko-extractor tool to ensure a clean text format without non-textual markers.
- Provided in a structured CSV format derived from the original Aozora Bunko index for easy ingestion.